Synthesis and pharmacological validation of a novel series of non-steroidal FXR agonists

Article abstract of DOI:10.1016/j.bmcl.2010.06.084

To overcome the known liabilities of GW4064 a series of analogs were synthesized where the stilbene double bond is replaced by an oxymethylene or amino-methylene linker connecting a terminal benzoic acid with a substituted heteroaryl in the middle ring po

Full text of DOI:10.1016/j.bmcl.2010.06.084

Bioorganic & Medicinal Chemistry Letters 20 (2010) 4911–4917
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and pharmacological validation of a novel series of non-steroidal
FXR agonists
Ulrich Abel a, , Thomas Schlüter a, Andreas Schulz a, Eva Hambruch a, Christoph Steeneck a,
a
a
a
a
b
´
Martin Hornberger , Thomas Hoffmann , Sanja Perovic-Ottstadt , Olaf Kinzel , Michael Burnet ,
Ulrich Deuschle a, Claus Kremoser a,
*
a Phenex Pharmaceuticals AG, Waldhofer Straße 104, 69123 Heidelberg, Germany
b Synovo GmbH, Paul-Ehrlich Straße 15, 72076 Tübingen, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
To overcome the known liabilities of GW4064 a series of analogs were synthesized where the stilbene
double bond is replaced by an oxymethylene or amino-methylene linker connecting a terminal benzoic
acid with a substituted heteroaryl in the middle ring position. As a result we discovered compounds with
increased potency in vitro that cause dose-dependent reduction of plasma triglycerides and cholesterol in
db/db mice down to 2 Â 1 mg/kg/day upon oral administration.
Received 20 April 2010
Revised 10 June 2010
Accepted 11 June 2010
Available online 19 June 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
Farnesoid X receptor
FXR agonist
GW4064
db/db mice
Metabolic syndrome
The Farnesoid X Receptor (FXR) is a member of the nuclear hor-
mone receptor superfamily and senses bile acid levels as its endog-
enous ligands.1–3 FXR is mainly expressed in tissues which are
exposed to bile acids such as the entire gastrointestinal tract, the
liver and the gallbladder. FXR mRNA can also be found in adrenals,
kidneys, adipose tissue, and possibly the pancreas.4,5 Activation of
FXR by derivatives of natural bile acid ligands such as 6-ethyl-che-
nodeoxycholic acid (6-ECDCA) or by synthetic non-steroidal ago-
nists like GW4064 results in beneficial metabolic effects such as
glucose lowering, insulin sensitisation, triglyceride, and cholesterol
lowering6–8 in various rodent models. Moreover, FXR agonists
seem to have specific hepatoprotective functions by preventing
lipid accumulation and reducing fibrosis and inflammation in the
liver, specifically. Beyond the effects provided by its liver-specific
functions, FXR controls the expression of a key cytokine, FGF-15/
19 in the intestine, specifically the ileum.9 FGF-19 has insulin sen-
sitizing, body weight lowering and lipid lowering effects, thereby
extending the action of FXR agonists indirectly to all tissues which
are generally responsive to FGF-15/19.10 Recently, the release of
positive data from two phase II studies using 6-ECDCA (INT-747),
a bile acid derivative and orally available FXR agonist, in patients
with Primary Biliary Cirrhosis (PBC), a chronic inflammatory chole-
static condition in the liver, and in Type II diabetics with non-alco-
holic fatty liver disease (NAFLD) validated the clinical utility of FXR
agonists in these metabolic and hepatic indications. 6-ECDCA
administration to Type II diabetic patients for 6 weeks improved
insulin sensitivity and lowered body weight in these individuals.11
However, high doses of 6-ECDCA seems to be associated with the
induction of pruritus in both PBC patients and healthy volunteers
(phase I study from the same group of investigators12). These po-
tential adverse effects of higher doses of 6-ECDCA may limit the
clinical utility of this drug, fostering the need for further FXR ago-
nists with potent in vivo activity.
GW4064 is the first synthetic non-steroidal FXR agonist de-
scribed in the literature.13 It demonstrates potent FXR binding
and activation in biochemical and cellular in vitro assays as well
as pharmacological effects in different rodent models of metabolic
and other diseases. However, its suitability as a drug for FXR tar-
geted pharmacotherapy is questionable given that this compound
harbors a trans stilbene moiety as a potential toxicophore and gi-
ven its photolability.14 Various groups have tried to overcome
the liabilities associated with this prototypical non-steroidal FXR
agonist and have generated derivatives of GW4064 without the
trans stilbene moiety.15,16 Most of these approaches, however,
did not provide compounds with improved potency at FXR.
Using GW4064 as a starting point, we have modified its struc-
ture to increase polarity, improve bioavailability, and potentially
* Corresponding author. Tel.: +49 6221 65282 0; fax: +49 6221 65282 10.
E-mail address: claus.kremoser@phenex-pharma.com (C. Kremoser).
Present address: Merz Pharma GmbH & Co. KGaA, Eckenheimer Landstraße 100,
60318 Frankfurt am Main, Germany.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.06.084

4912
U. Abel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4911–4917
also increase potency at FXR. Here we report the synthesis and
pharmacological evaluation of a novel series of FXR agonists
wherein the core features of GW4064 are mostly maintained and
the trans stilbene moiety of GW4064 is replaced by 3- and 4-ben-
zoic acid moieties linked to a substituted pyridine in the middle
ring position by oxymethylene or amino-methylene linkers in both
possible orientations. For those cases where the –O– or –NH– or –
NMe– was the attachment point to the middle ring, which is linked
to the isoxazole by another oxymethylene linker, we replaced the
3-Cl-phenyl from the GW4064 by different –CF3 substituted pyri-
dines in order to reduce the potential formation of quinoidal struc-
tures. Table 1 shows the compounds which have been synthesized
and their activities in a biochemical FXR FRET assay, a cellular
Gal4-LBD reporter gene assay (M1H) and a full length FXR direct
reporter (DR) assay.
Schemes 1–3 provide an overview on the synthetic strategies
chosen for making various derivatives within this structural class
of close GW4064 analogs. The pyridine-3-yl-containing analogs
are prepared according to Scheme 1. Briefly, commercially avail-
able 6-methyl-5-nitropyridin-2-ol (1a) or the corresponding tri-
fluoro-methyl analog (1b) were alkylated with isoxazole building
blocks 2a or 2b via a Mitsunobu reaction using DIAD. The nitro
group of the intermediates were reduced with zinc in acetic acid
to afford the common intermediates 3a–c. Standard derivatization
such as amide bond formation with benzoyl chlorides or reductive
alkylation with benzaldehydes, subsequent alkylation of the amide
NH or the secondary amine with MeI or EtI and final ester hydro-
lysis gave analogs 5–8, 12–15, 17, 19, 22, and 24. The hydroxyethyl
analog 18 was prepared in a similar fashion via a reductive alkyl-
ation of 3b with benzaldehyde 4b followed by a final hydroxyl
deprotection. The analog 9 was prepared by a Mitsunobu reaction
of 4-methyl-5-nitropyridin-2-ol with 2a whereas 10 and 11 were
prepared by reaction of deprotonated 2a with 2-chloro-4-methyl-
quinoline and 2-chloro-6-methyl-4-(trifluoro-methyl)-pyridine,
respectively.
Scheme 2 shows the synthesis of analogs with modified acidic
moieties. Acyl sulfonamides 16 and 20 are obtained from 13 and
15, respectively, by reaction with EDCI and methane sulfonamide,
while analog 21 is obtained in a two step sequence, first activation
of the carboxylic acid as an ONSu ester and subsequent reaction
with O-phospho-ethanolamine in the presence of sodium bicar-
bonate as base. For sulfonic acid derivative 23 the best result was
obtained when taurine was coupled to the carboxylic acid using
the phosphonium based coupling reagent HATU. The two pyri-
dine-2-yl-containing analogs 27 and 28 were prepared as follows
(Scheme 2): 2,3-dichloro-5-nitropyridine was reacted by nucleo-
philic substitution of the chlorine in 2-position with methyl 3-
(aminomethyl)-benzoate or methyl 4-(aminomethyl) benzoate,
respectively, followed by diazotation and iodination at the 5-posi-
tion. A CuI/phenanthroline catalyzed cross coupling reaction of
intermediates 26a or 26b with 2a, followed by ester hydrolysis
afforded compounds 27 and 28.
Scheme 3 summarizes the synthesis of analogs containing a
phenolether linker element (35–39) and analogs 33 and 34, con-
taining inverted amino-methylene linkers. Commercially available
tetrahydropyridone derivative 29 was converted to the corre-
sponding 2-hydroxy-pyridine by bromination with NBS and elimi-
nation of HBr. This intermediate is alkylated with 2b via a
Table 1
Overview of compounds made as FXR active GW4064 analogs
R1
O
O
N
11
GW4064
O
N
Cl
O
N
F
N
5–8; 12–39
Y2
R2
O
-O
O
N
F
Cl
O
O
Y1
N+
O
Cl
O
Cl
O
N
X1
Cl
F
Cl
Y3
N
O
N
Cl
Cl
X2
HO
Cl
Cl
R3
Cl
9
10
Compd
R1
R2
R3
X1
X2
Y1
Y2
Y3
FRET (nM)
Eff (%)
M1H (nM)
Eff (%)
DR (nM)
Eff (%)
GW4064
5
6
7
8
See extra structure
p-COOH
p-COOH
p-COOH
m-COOH
See extra structure
See extra structure
See extra structure
p-COOH
p-COOH
p-(CH2)2–COOH
m-COOH
p-CONH–SO2–CH3
p-OCH2–COOH
p-O(CH2)2–OH
m-COOH
m-CONH–SO2–CH3
m CONH–(CH2)2–OPO3H2
p-COOH
p-CONH–(CH2)2–SO3H
p-COOH
p-COOH
20
235
452
365
433
89
147
225
162
22
17
12
40
35
164
3
48
36
12
100a
101
109
95
110
101
104
98
97
97
96
102
105
90
35
316
474
2030
571
1000
536
Inactive
398
90
100a
36
21
21
35
58
40
Inactive
48
75
55
89
64
69
43
79
59
48
81
57
54
70
30
208
385
1190
537
1140
878
Inactive
101
15
16
6
11
19
155
2
20
216
6
925
113
120
61
100a
63
48
49
55
76
68
Inactive
45
57
66
58
67
59
59
43
72
61
60
49
43
51
54
49
57
33
41
42
49
33
i-Pr
i-Pr
i-Pr
i-Pr
CF3
CF3
CH3
CF3
CO
CO
CO
CO
N–CH3
NH
N–CH3
N–CH3
CH
CH
CH
CH
CH
CH
CH
CH
N
N
N
N
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
27
28
33
34
35
36
37
38
39
i-Pr
i-Pr
i-Pr
i-Pr
i-Pr
i-Pr
i-Pr
c-Pr
i-Pr
i-Pr
c-Pr
c-Pr
c-Pr
i-Pr
i-Pr
c-Pr
c-Pr
c-Pr
c-Pr
c-Pr
c-Pr
c-Pr
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
Cl
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
NH
NH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
N
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
N
N
N
N
N
N
N
N
N
N
N
N
N
CH
CH
N
N
N
N
N
N
N
N–CH3
N–CH3
N–CH3
N-CH3
N–CH3
N–CH3
N-CH3
N–CH3
N–CH3
N–CH3
N–CH3
N–CH2CH3
N–CH3
N–CH3
CH–CH3
CH–CH3
CH2
35
80
221
214
1150
9
279
457
30
1420
180
107
100
333
208
1710
151
95
87
102
110
114
109
105
94
103
105
99
102
91
98
86
102
89
14
194
242
190
101
179
434
43
76
155
352
Cl
m-COOH
p-COOH
p-COOH
o-COOH
m-COOH
p-COOH
m-COOH
p-COOH
N
42
59
55
63
62
59
50
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CH
CH
CH
CH
CH
CH
CH
202
182
611
33
38
58
N–CH3
O
O
O
O
O
CH2
CH2
CH–CH3
CH–CH3
107
189
70
611
a
Efficacy is set as 100%. The experimental details can be found in the supplementary data.

U. Abel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4911–4917
4913
R1
2a; R1 = i-Pr
2b; R1 = c-Pr
O
N
R1
O
N
HO
OH
a), b)
O
Cl
Cl
Cl
N
N
Cl
O2N
H2N
R2
2 = Me
O
R2
3a: R2 = Me; R1 = i-Pr
1a:
1b:
R
R
4a;
R = CH2CO2Et
2 = CF3
3b: R2 = CF3; R1 = i-Pr
4b; R = (CH2)2OTBDMS
OR
3c:
R
2 = CF3; R1 = c-Pr
c), d), e)
h), d), e)
c), d), e)
k), g), l)
17
5, 8
3a
7
3b
2a
k), g), p)
c), e)
f), e)
18
6
19
22
24
m)
n)
i), d), e)
i), j), e)
3b
12
9
3c
f), g), e)
h), g), e)
10
11
13, 14
15
o)
Scheme 1. For advanced starting materials see Ref. [21], reagents and conditions: (a) 2, PPh3, C6H6 or PhCH3, DIAD, rt, 12 h; 54–88%; (b) Zn, AcOH, MeOH, rt, 2 h, 75–95%;
(c) methyl 4-(chlorocarbonyl) benzoate (for 5) or methyl 3-(chlorocarbonyl)benzoate (for 8), TEA, CH2Cl2, rt, 2 h, 31–65%; (d) NaH, MeI, THF, rt, 17 h; 56–79%; e) NaOH, H2O,
50 °C, 3 h; 54–100%; (f) methyl 4-formylbenzoate (for 13) or methyl 3-(4-formyl-phenyl)propanoate (for 14), NaBH(OAc)3, AcOH, DCE, rt, 24 h, 49%; (g) paraformaldehyde,
NaBH(OAc)3, AcOH, DCE, rt, 24 h, 59–70%; (h) methyl 3-formylbenzoate, THF, BF3ÁEt2O, rt, 12 h, then Na(CN)BH3, rt, 1–2 h, 89–95%; (i) methyl 4-formylbenzoate, THF,
BF3ÁEt2O, rt, 12 h, then Na(CN)BH3, rt, 4 h, 70%; (j) EtI, NaH, DMF, 0 °C, 2 h; (k) 4a (for 17) or 4b (for 18), THF, BF3ÁEt2O, rt, 12 h, then Na(CN)BH3, rt, 1 h, 53–70%; (l) LiOH, THF,
H2O, rt, 12 h, 83–92%; (m) 4-methyl-5-nitropyridin-2-ol, PPh3, DIAD, C6H6, rt, 12 h, 56%; (n) 2-chloro-4-methylquinoline, NaH, THF, rt, 12 h, 70%; (o) 2-chloro-6-methyl-4-
(trifluoromethyl)pyridine, NaH, THF, rt, 12 h, 55%; (p) TBAF, THF, rt, 5 h, 75%.
21: R1 = i-Pr ;R3 = 3-CONH(CH2)2OPO3H2
b, c)
R1
Cl
15: R1 = i-Pr; R3 = 3-CO2H
O
N
13: R1 = i-Pr; R3 = 4-CO2H
a)
20: R1 =i -Pr; R3 = 3-CONHSO2Me
16: R1 = i-Pr; R3 = 4-CONHSO2Me
O
a)
Cl
N
N
R3
CF3
22: R1 = c-Pr; R3 = 4-CO2H
d)
23: R1 = c-Pr; R3 = 4-CONH(CH2)2SO2H
X
NO2
N
N
i), j)
e), f)
27
28
N
Cl
R3
Cl
Cl
25a: 3-COOMe; X = NO2
25b: 4-COOMe; X = NO2
26a: 3-COOMe; X = I
26b: 4-COOMe; X = I
g), h)
Scheme 2. Reagents and conditions: (a) methane sulfonamide, EDCI, DMAP, CH2Cl2, rt, 12 h, 50–60%; (b) HOSu, DCC, EtOAc, rt, 12 h, 93%; (c) O-phospho-ethanolamine,
NaHCO3, THF, H2O, rt, 48 h, 17%; (d) HATU, DMF, 0 °C, 30 min, then taurine, DIEA, rt, 18 h, 50%; (e) methyl 3-(aminomethyl) benzoate hydrochloride or methyl
4-(aminomethyl)benzoate hydrochloride, DIPEA, i-PrOH, 60 °C, 1 h, 92–96%; (f) NaH, MeI, DMF, rt, 2 h, 78–96%; (g) Na2S2O4, MeOH, H2O, 90 °C, 1 h, 36–39%; (h)
isoamylnitrite; CH2I2, cat. HI, rt, 2 h, 44–47%; (i) 3.5 equiv 3a, cat. Cu, cat. 1,10-phenanthroline, Cs2CO3, PhCH3, 120 °C, 14 h, 65–73%; (j) LiOH, THF, MeOH, H2O, rt, 6 h, 73–79%.
Mitsunobu reaction to afford common intermediate 30. After
reduction of 30 to the corresponding hydroxymethyl analog 31a
with LAH compounds 35–37 were obtained by Mitsunobu alkyl-
ation with o-, m-, and p-methyl benzoates and final ester hydroly-
sis. Analogs 38 and 39 were obtained by alkylation of methyl 3- or
4-hydroxybenzoate with chloride 31c and final ester hydrolysis.
Chloride 31c in turn was prepared from 30 by transformation of
the ethyl ester into the Weinreb amide, reaction with MeMgBr,
reduction of the ketone to the alcohol and ensuing chlorination.
Analogs 33 and 34, containing inverted amino-methylene linkers,
were prepared by forming first the Schiff-base of methyl 4-amino-
benzoate with aldehyde 31d, followed by addition of a methyl
group by reaction with MeLi, then either direct hydrolysis of the
ester (33) or methylation of the secondary amine and final ester
hydrolysis (34).
Modification of the original isopropyl at the 5-position of the
isoxazole into a cyclopropyl results in an about two to fourfold
increase in FXR activation potency (compare 13 to 22 and 15 to
19). Notably, derivatives that just contain the middle heteroaryl
ring either as a bicyclic structure such as in 10 or as substituted
pyridyl like in 9 or 11 maintain FXR agonistic capabilities as the
FRET biochemical assay data indicate. This suggests that a core

4914
U. Abel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4911–4917
O
N
30: R = CO2Et
O
c)
O
e)-g)
h), i)
a), b)
31a: R = CH2OH
31b: R = C(O)Me
31c: R = CHClCH3
31d: R = CHO
Cl
NH
N
EtOOC
d)
Cl
R
29
CF3
CF3
35
36
37
l), k)
j), k)
38
39
31c
31a
31d
O
N
O
o)
33
34
H
Cl
m), n)
N
N
p), o)
Cl
CF3
MeOOC
32
Scheme 3. Reagents and conditions: (a) NBS, CCl4, reflux 18 h; (b) 2b, PPh3, DIAD, THF, rt 1.5 h, 74% (two steps); (c) LAH, THF, rt, 2 h, 86%; (d) MnO2, CHCl3, reflux, 3 h, 92%; (e)
LiOH, MeOH, H2O, rt, 12 h, 100%; (f) N,O-dimethylhydroxylamine hydrochloride, HATU, DIPEA, DMF, rt, 12 h, 82%; g) MeMgBr, THF, À78 °C to rt, 2.5 h, 79%; (h) LAH, THF, 0 °C,
2 h, 94%; (i) SOCl2, CHCl3, 0 °C, 2 h, 92%; (j) methyl hydroxybenzoate (o-, m- or p-), PPh3, DIAD, THF, rt, 12 h, 60–65%; (k) LiOH, MeOH or THF, H2O, reflux, 5 h, 60–90%; (l)
methyl 3-hydroxybenzoate or methyl 4-hydroxybenzoate, NaH, DMF, 100 °C, 12 h, 22–38%; (m) methyl 4-aminobenzoate, PPTS, PhCH3, Dean–Stark, 16 h; (n) MeLi, THF, rt,
1.5 h, 56%; (o) LiOH, THF, H2O, rt, 12 h, 36–48%; (p) NaH, THF, MeI, 0 °C to rt, 2 h, 48%.
structure containing the 2,6-dichlorophenyl-isoxazole moiety con-
tributes a major part to the binding as well as to the activation
properties of this type of ligands. While omitting the linker and
aryl-acid part results in an only minor loss of FXR cell-free activity
the loss of potency in the cellular assays is more dramatic.
The substituents R2 contribute potency with a –CF3 being the
best amongst –Cl, –CF3, and –CH3. Halogens have not been tried
in the case of the pyridin-3-yl because the proximity to the pyri-
dine nitrogen may cause an undesirable instability of such substi-
tuted heteroaryls.
The two pyridin-2-yl derivatives (27, 28) have chlorines in the
R2 position of the middle ring and therefore resemble the original
substitution pattern of GW4064 except for the different linker and
the pyridine-N. Compared to the pyridine-3-yl compounds these
two examples are significantly less potent in all three assays and
were not further explored.
A strong impact on the FXR activity of these isoxazole deriva-
tives comes from the configuration of the two-atom spacer be-
tween the middle (hetero)aryl and the COOH bearing terminal
phenyl. The crystal structure of GW4064 suggests to replace the
stilbene double bond by a linker of similar length with the addi-
tional constraint that the slightly staggered coplanar conformation
of the middle aryl and the terminal benzoic acid might be kept in
order to maintain FXR agonism and potency. We synthesized
(methyl)-amino-methylene and oxymethylene linkers in both ori-
entations between the middle heteroaryl and the terminal benzoic
acid except for the oxymethylene with the –O– attached to the
middle heteroaryl in order to avoid potential benzoquinone-like
structure formation in vivo. The N-methyl substituted amino-
methylene linker with nitrogen attached to the middle aryl yielded
by far the best results. Omitting the methyl substituent at the
nitrogen results in an appr. 20-fold loss of activity whereas
replacement of N-methyl by ethyl results in decreased FXR activity,
suggesting that there is no larger lipophilic pocket to accommodate
the N-alkyl substituent (compare 12, 13, 22, and 24, respectively).
Changing the orientation of the heteroatom in the linker decreases
potency by about 5- to 10-fold (compare 22 vs 34).
lar assays. When these benzoic acids are changed into –COOH bio-
isosteres with similar acidity such as the acylsulfonamides 16 and
20, this correlation is still maintained. However, compounds which
have permanently charged acidic functions with increased polari-
ties such as 21 or 23 display a strong disparity between the FRET
results which indicate still good binding to FXR and the M1H and
DR data demonstrating a rather poor cellular activity, probably
due to their diminished membrane permeability. Just increasing
the distance between the phenyl and the –COOÀ moiety as exem-
plified in 14 and 17 as opposed to 13 neither has a strong impact
on the potency in the FRET assay nor in the cellular assay.
Molecular modeling studies suggest that the prolonged distance
to the acidic function should be incompatible with the formation of
the known ionic interaction to Arg331 as described in Refs. 14 and
15 Pellicciari et al.17 suggest that there is a ‘hole’ in the FXR ligand
binding pocket structure which marks an exit vector for substitu-
ents ranging out from the p-position of the terminal aryl passing
along the Arg331 that forms an important salt bridge with the ben-
zoic acid –COOÀ of the ligand towards the aqueous surface of the
receptor. If compounds 14, 21, and 23 really reach out into the sol-
vent surrounding of the FXR protein, then the only minor differ-
ences in the biochemical assay potency might imply that the loss
of binding enthalpy (see Fig. 1) by losing the ionic interaction to
Arg331 might be compensated by the gain in H2O solvatisation en-
thalpy. Docking studies with ligands 13 and 23 suggest, however,
À
that the —SO3 group orients towards Arg264 resulting in an ionic
interaction in addition to maintaining an H-bond interaction be-
tween the amide –C@O and Arg331 as well as forming a new one
to Ser342. Since most docking programs cannot precisely account
for solvation enthalpy effects it remains unclear whether in reality
the solvation effect or the formation of a new charged interaction
prevails.
Since the active isoxazole derivatives from this series should
mimic the natural ligand bile acids at FXR we speculated that trans-
fection of the bile acid transporter NTCP (= Na-taurocholate cotrans-
port protein) that actively imports taurine conjugated bile acids into
hepatocytes18 might also accept the taurine conjugate 23 as a sub-
strate which could alter 23 potentially into a cellular and possibly
in vivo active compound. Indeed, upon transfection of NTCP into
the HEK 293 cells which are used for the M1H assay, the EC50 value
decreased from 1417 nM down to 36 nM, comparable to the 30 nM
Considerable attention should be given to the nature and the
position of the acid moiety at the terminal phenyl ring. All deriva-
tives that contain a carboxylic acid at the terminal phenyl show a
consistent correlation between activities in biochemical and cellu-

U. Abel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4911–4917
4915
Figure 1. Solvent surface of FXR colored by interpolated charge with the docked compounds 13 (A, C) and 23 (B, D), respectively.
of 22. This shows that the taurine amide of 22 exerts its good FXR po-
tency in a cellular environment, when the membrane permeability
barrier is overcome by an active transport mechanism.
detected (see Tables 2A and 2B). Interestingly, 23, which shows
poor cell permeability in cellular assays was found at high levels
in the liver and in bile fluid. Compound 23, the taurine amide of
22, decomposed to 22 only to a minor extent, whereas 22 was con-
verted into 23 to a large extent upon liver passage. The unusual
taurine amidation as one form of Phase II metabolism resembles
the taurine conjugation of bile acids. This suggests that these isox-
azole derivatives are not only potent FXR agonists but also mimic
bile acids as the natural FXR ligands with regard to being recog-
nized by at least the two enzymes involved in bile acid–taurine
conjugation and the transporters involved in subsequent transport.
Since NTCP, as the major hepatic import pump for conjugated bile
acids accepts 23 as a substrate and 22 is found to be converted into
23 in vivo one may assume an enterohepatic circulation of the tau-
rine amidated compounds.19
All compounds described in Schemes 1–3, were tested in a
broad human nuclear receptor M1H assay panel to check for off-
target activity within this target class (see Table 1 in supplemen-
tary data). Compounds 13 and 22 were found to have the lowest
off-target activity with basically no efficient activation of any other
nuclear receptor except for FXR and PXR at appr. 100-fold less po-
tency compared to their FXR activities. Compound 19, although the
most potent in all assays used, shows minor activity at the Estro-
gen Receptors
a and b, PPARa, and PPARc and was for this reason
not considered further for in vivo testing.
Compounds 13, 23, and 22 were subject to pharmaco-kinetic
and pharmacodynamic studies in mouse. Surprisingly, we could
find only minute amounts of the compounds in plasma, indepen-
dent of the route of administration, po or iv. However, when look-
ing for parent compound and main metabolite levels in the liver
and bile fluid, very high concentrations of the parent compound
as well as of the direct glucuronides and taurine amides could be
The other relevant key metabolites of 13 and 22, besides minor
amounts of N-demethylation products, were the respective glucu-
ronides. These were found to be acyl-glucuronides since digestion
of bile or bile extracts of animals treated with 13 or 22 with glucu-
ronidase yielded the unmodified parent compounds and the COOH

4916
U. Abel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4911–4917
Table 2A
Plasma pharmacokinetics of the compounds 13, 22, and 23
Compd 13
Compd 22
Compd 23
5 (mg/kg) po
1 (mg/kg) iv
5 (mg/kg) po
1 (mg/kg) iv
5 (mg/kg) po
1 (mg/kg) iv
5 (min)a
57
92
420
267
14
5260
6253
16.2
1617
1598
361
122
0
427
283
531
370
745
437
325
398
94
1
3
6
12
72
59
3
19
10 (min)a
30 (min)a
90 (min)a
240 (min)a
Vd (mL/kg)
CL (mL/h/kg)
Bioavailability (%)
52
0
8
931
1010
5200
4075
16.1
1243
717
n.a.
196,931
5.8
31,691
11,385
a
Plasma (parent compd only). All values in ng/mL. n.a. = data not available.
Table 2B
Liver and gall content of the compounds 13, 22, and 23 and their key metabolites
Compd Dose/Route
Livera (carboxylic acid) Livera (glucuronide) Livera (taurine amide) Bilea (carboxylic acid) Bilea (glucuronide) Bilea (taurine amide)
13
13
22
22
23
23
5 (mg/kg) po 1512
2872
0
172
17
30
7
372
134
0
0
262
350
1697
472
1446
173
32
18,928
4261
2807
150
1055
169
22,443
11,927
47,638
10,013
32,442
24,151
1 (mg/kg) iv
5 (mg/kg) po
1 (mg/kg) iv
5 (mg/kg) po
1 (mg/kg) iv
262
257
23
6
8
12
a
Time after dosing 240 min. All values in ng/mL.
moiety presents the only functional group of the parent com-
pounds amenable for direct glucuronidation.
To test for pharmacodynamic effects with regard to lipid lower-
ing and for potential compound accumulation over prolonged
exposure, we used the db/db mouse, an accepted model for FXR-
mediated effects. Compounds 13 and 22 were administered at
doses of 1, 5, and 25 mg/kg b.i.d. and compared with the effects
on total plasma triglycerides (TG) and total plasma cholesterol
(TC) of 6-ECDCA (see Fig. 2). All three compounds, 6-ECDCA, 13,
and 22, showed TG and TC lowering effects that were significant
for most doses tested (p <0.01 for 13 and 6-ECDCA at 1 mg/kg
b.i.d. and p <0.001 for all three compounds at all other doses for
TG lowering; p <0.001 for 13 and 22 at 2 Â 5 mg/kg b.i.d. and all
higher doses and for 6-ECDCA at 25 mg/kg b.i.d. for TC lowering).
All three FXR agonists tested showed a clear linear dose response
relationship with regards to TG and TC lowering, however, 22
was found to be the most effective in terms of lipid lowering effects
The fact that all three compounds were detected at double digit
micromolar concentrations as glucuronides and/or taurine amides
in the bile suggests a model wherein these molecules are biologi-
cally recognized as bile acids, are taken up by active transport in
the intestine and undergo a high first pass liver extraction (most
likely also due to NTCP or similar transporter mediated active up-
take), phase II conjugation into acyl-glucuronides and taurine
amides, respectively, and subsequent export into bile. The taurine
conjugates are potent FXR agonists themselves and are likely to be
subject to reuptake in the intestine, thereby providing a mecha-
nism for sustained pharmacodynamic action. The inherent danger
of multiple rounds of enterohepatic cycling lies in accumulation of
compound upon regular daily dosing.20
Figure 2. C57BL/KSK-mLepdb/mLepdb mice were treated (12 days) with the compounds 13, 22, or 6-ECDCA ( p <0.05, p <0.01,
**
*
***
p <0.001).

U. Abel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4911–4917
4917
(TC mean for 22 at 1 mg/kg b.i.d. = 97.2 mg/dl vs 121.0 mg/dl for
1 mg/kg b.i.d. 6-ECDCA, p <0.05).
mediated FXR activation in multiple clinically relevant disease
models.
With respect to compound accumulation after multiple dosing
the differences are striking (Table 2 in supplementary data):
Whereas all three compounds tested show comparable levels of
low plasma but very high liver concentrations at 4 h after the last
administration, suggesting a high first pass effect for these com-
pounds, organ concentrations at 24 h after the last administration
differ dramatically between the isoxazole derivatives and 6-ECD-
CA. For 13, 22, and its metabolite 23 only minute amounts were
detectable in either plasma or liver whereas 6-ECDCA showed
nearly unchanged levels in plasma and liver, indicative of either
highly efficient enterohepatic circulation, unusual stability, or stor-
age in the liver.
The absence of 22 and 23 24 h after the last administration de-
spite the proposed entereohepatic cycling of the taurine amide
may be explained by the fact that the taurine amide is only one
of a few key metabolites including the N-demethylated species
and the glucuronide. The glucuronide of 22 may not be accepted
by the bile acid transporters which would result ultimately in fecal
loss given that cleavage of glucuronide by intestinal bacteria is not
complete. Despite, the relatively rapid removal of the isoxazoles 13
and 22 from the hepatobilliary system, they show pharmacody-
namic effects in the db/db mouse comparable or better than those
of 6-ECDCA which has a dramatically longer duration of exposure.
In summary we have made a series of potent FXR agonists that
mimic the structure of the parent compound GW4064 but are de-
void of the stilbene moiety. A key functional feature of this series is
either the free terminal benzoic acid or an amide conjugate thereof
that can extend the position of the acidic moiety by two atoms
without significantly losing potency in biochemical or cellular
FXR assays. When these extended derivatives such as 14, 17, 21,
or 23 are docked into the crystal structure of GW4064-bound
FXR, the structurally related elements adopt very similar positions
to those of GW4064. The terminal acid, however, which is believed
to form an ion bridge to Arg331 in the GW4064-FXR co-crystal
might change its ion bridge partner to Arg264.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2010.06.084.
References and notes
1. Parks, D. J.; Blanchard, S. G.; Bledsoe, R. K.; Chandra, G.; Consler, T. G.; Kliewer,
S. A.; Stimmel, J. B.; Willson, T. M.; Zavacki, A. M.; Moore, D. D.; Lehmann, J. M.
Science 1999, 284, 1365.
2. Makishima, M.; Okamoto, A. Y.; Repa, J. J.; Tu, H.; Learned, R. M.; Luk, A.; Hull,
M. V.; Lustig, K. D.; Mangelsdorf, D. J.; Shan, B. Science 1999, 284, 1362.
3. Forman, B. M.; Goode, E.; Chen, J.; Oro, A. E.; Bradley, D. J.; Perlmann, T.;
Noonan, D. J.; Burka, L. T.; McMorris, T.; Lamph, W. W.; Evans, R. M.;
Weinberger, C. Cell 1995, 81, 687.
4. Higashiyama, H.; Kinoshita, M.; Asano, S. Acta Histochem. 2008, 110, 86.
5. Bookout, A. L.; Jeong, Y.; Downes, M.; Yu, R.; Evans, R. M.; Mangelsdorf, D. J.
http://www.nursa.org/10.1621/datasets.02001.
6. Cariou, B.; van Harmelen, K.; Duran-Sandoval, D.; van Dijk, T. H.; Grefhorst, A.;
Abdelkarim, M.; Caron, S.; Torpier, G.; Fruchart, J. C.; Gonzalez, F. J.; Kuipers, F.;
Staels, B. J. Biol. Chem. 2006, 281, 11039.
7. Zhang, Y.; Lee, F. Y.; Barrera, G.; Lee, H.; Vales, C.; Gonzalez, F. J.; Willson, T. M.;
Edwards, P. A. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 1006.
8. Ma, K.; Saha, P. K.; Chan, L.; Moore, D. D. J. Clin. Invest. 2006, 116, 1102.
9. Inagaki, T.; Choi, M.; Moschetta, A.; Peng, L.; Cummins, C. L.; McDonald, J. G.;
Luo, G.; Jones, S. A.; Goodwin, B.; Richardson, J. A.; Gerard, R. D.; Repa, J. J.;
Mangelsdorf, D. J.; Kliewer, S. A. Cell Metab. 2005, 2, 217.
10. Jones, S. Mol. Pharmacol. 2008, 5, 42.
11. Sanyal, A. J.; Mudaliar, S.; Henry, R. R.; Marschall, H. U.; Morrow, L.; Sciacca, C.
I.; Dillon, P.; Clopton, P.; Kipnes, M.; Shapiro, D. Hepatology 2009, 50, 389A.
12. Shapiro, D.; March, B.; Eliot, L.; duBois, R.; Pruzanski, M. Hepatology 2009, 50,
770A.
13. Maloney, P. R.; Parks, D. J.; Haffner, C. D.; Fivush, A. M.; Chandra, G.; Plunket, K.
D.; Creech, K. L.; Moore, L. B.; Wilson, J. G.; Lewis, M. C.; Jones, S. A.; Willson, T.
M. J. Med. Chem. 2000, 43, 2971.
14. (a) Bass, J. Y.; Caldwell, R. D.; Caravella, J. A.; Chen, L.; Creech, K. L.; Deaton, D.
N.; Madauss, K. P.; Marr, H. B.; McFadyen, R. B.; Miller, A. B.; Parks, D. J.; Todd,
D.; Williams, S. P.; Wisely, G. B. Bioorg. Med. Chem. Lett. 2009, 19, 2969; (b) The
photolability of GW4064 was compared to the photolability of 22 and 37, see
Supplementary data.
15. Akwabi-Ameyaw, A.; Bass, J. Y.; Caldwell, R. D.; Caravella, J. A.; Chen, L.; Creech,
K. L.; Deaton, D. N.; Jones, S. A.; Kaldor, I.; Liu, Y.; Madauss, K. P.; Marr, H. B.;
McFadyen, R. B.; Miller, A. B.; Iii, F. N.; Parks, D. J.; Spearing, P. K.; Todd, D.;
Williams, S. P.; Wisely, G. B. Bioorg. Med. Chem. Lett. 2008, 18, 4339.
16. Feng, S.; Yang, M.; Zhang, Z.; Wang, Z.; Hong, D.; Richter, H.; Benson, G. M.;
Bleicher, K.; Grether, U.; Martin, R. E.; Plancher, J. M.; Kuhn, B.; Rudolph, M. G.;
Chen, L. Bioorg. Med. Chem. Lett. 2009, 19, 2595.
17. Pellicciari, R.; Gioiello, A.; Costantino, G.; Sadeghpour, B. M.; Rizzo, G.; Meyer,
U.; Parks, D. J.; Entrena-Guadix, A.; Fiorucci, S. J. Med. Chem. 2006, 49, 4208.
18. Hagenbuch, B.; Meier, P. J. J. Clin. Invest. 1994, 93, 1326.
19. Hofmann, A. F.; Hagey, L. R. Cell. Mol. Life Sci. 2008, 65, 2461.
20. Roberts, M. S.; Magnusson, B. M.; Burczynski, F. J.; Weiss, M. Clin.
Pharmacokinet. 2002, 41, 751.
21. Literature references for advanced starting materials. (4a) Kiec´-Kononowicz,
K.; Karolak-Wojciechowska, J.; Michalak, B.; Pekala, E.; Schumacher, B.; Müller,
C. E. Eur. J. Med. Chem. 2004, 39, 205.; (4b) Larsen, S. D.; May, P. D.; Romines, K.;
Schnute, M. E.; Tanis, S. P.; Nieman, J. A.; Anderson, D. J.; Kim, E. J.; Turner, S. R.
PCT Int. Appl. 2003, WO 2003059878.; (1a) Kremoser, C.; Deuschle, U.; Abel, U.;
Schulz, A. PCT Int. Appl. 2008, WO 2008025540.; (1b) Abel, U.; Kremoser, C.;
PCT Int. Appl. 2009, WO 2009149795.; (2a, 2b) Ref. 13; (29) Brown, S. M.;
Bowden, M. C.; Parsons, T. J.; McNeilly, P.; de Fraine, P. J. Org. Process Res. Dev.
1997, 1, 370.
When tested in db/db mice, compounds 13 and 22 showed a
linear dose-dependent reduction in total plasma triglycerides and
total plasma cholesterol. Surprisingly the non-steroidal FXR ago-
nists 13 and 22 are metabolically transformed (taurine and glucu-
ronic acid conjugation) in a way that resembles that of natural bile
acids. The key metabolite of 22, 23 is a direct ligand of FXR, and a
substrate of NTCP as observed by gain-of-function studies. Never-
theless, compared to 6-ECDCA which was present at high levels
in liver 24 h after the last administration, neither 22 nor 23 were
present at relevant concentrations in plasma or liver at this time-
point. Thus, the risk of sustained non-mechanism related effects
with the isoxazoles might be lower compared to 6-ECDA, while
the duration of desirable pharmacodynamic effects is clearly ade-
quate to modify lipid homeostasis in these mice.
Overall, these data suggest that the newly identified com-
pounds are useful tools for assessing the effect of small-molecule